Date: Tue, 03 Aug 2004 10:42:50 -0600 From: Scott Long <scottl@samsco.org> To: Evan Dower <evantd@hotmail.com> Cc: freebsd-current@freebsd.org Subject: Re: Stability? Message-ID: <410FC08A.7060702@samsco.org> In-Reply-To: <BAY8-F65Ezx0AEG2UYh0005e7c5@hotmail.com> References: <BAY8-F65Ezx0AEG2UYh0005e7c5@hotmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Evan Dower wrote: > % uptime > 9:23AM up 1 day, 15:42, 6 users, load averages: 0,98 1,19 1,25 > > With a recent cvsup (and all the rebuilding and installing involved) I > have noticed what seems to be an improvement in stability. I must have > missed any postings about it. Any, recent cvsup means about an hour > before this: > > % uname -a > FreeBSD lojak.washington.edu 5.2-CURRENT FreeBSD 5.2-CURRENT #0: Sun > Aug 1 16:06:05 PDT 2004 > root@lojak.washington.edu:/usr/obj/usr/src/sys/DEBUG i386 > > I noticed that my kernel has now started saying (in the dmesg): > WARNING: Kernel preemption is disabled, expect reduced performance. > I turned off normal preemption a few days ago in order to cover up the stability problems until the responsible parties (including myself) get them fixed. It is just a hack for now. > Of course, I have WITNESS turned on in my debug kernel so I already > expect reduced performance. Is this message suggesting that I put > FULL_PREEMPTION in my kernel? I was under the impression that it was > only useful for seeing where stuff broke. Is stuff so good now that it > will improve performance without significant effect on stability? Or > perhaps it is indicating that PREEMPTION is #undef'd? I didn't #undef it > so if this is the case it must have been changed upstream. You likely do not want to turn on FULL_PREEMPTION. If you want the old behavior, edit /sys/i386/include/param.h and reenable the PREEMPTION definition. Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?410FC08A.7060702>